Transformer sample selection method based on improved genetic algorithm

Abstract

The invention discloses a transformer sample selection method based on an improved genetic algorithm. The transformer sample selection method is concretely implemented according to the following steps: step 1, binary coding is performed on training samples and population initialization is performed, and the maximum number of iterations is set as T and the population size is set as N; step 2, the population is divided into a subpopulation of which the individuals are set as n after the step 1, the subpopulation is called a detection block, a population including great individuals is generated in the evolution process and called a development block, and the development block and the detection block are combined to select transformer samples; and step 3, after completion of the step 2, the multi-population genetic algorithm is improved and enhanced by using a cultural algorithm, the final individual of the maximum fitness is obtained under the corresponding evolution operation and transformer sample selection based on the improved genetic algorithm is completed. According to the transformer sample selection method based on the improved genetic algorithm, training sample optimal selection of the transformer is performed by using the multi-population genetic algorithm and the algorithm is improved by using the cultural algorithm so that the optimal sample can be acquired.

Description

technical field [0001] The invention belongs to the technical field of transformer monitoring methods, in particular to a transformer sample selection method based on an improved genetic algorithm. Background technique [0002] The transformer is one of the main equipment in the substation, and its fault diagnosis is helpful to improve the stability and reliability of the power grid operation. [0003] Nowadays, a large number of intelligent algorithms are used in transformer fault diagnosis and have achieved good results. It mainly combines the intelligent classification algorithm with oil chromatography to judge the operation status of the transformer. The more classic ones include BP neural network, fuzzy system, expert system, support vector machine, etc. At the same time, a large number of search algorithms are used in transformer fault diagnosis as optimization algorithms for classification algorithms, such as: more classic genetic algorithms, ant colony algorithms, a...

AI Technical Summary

Problems solved by technology

At the same time, a large number of search algorithms are used in transformer fault diagnosis as optimization algorithms for classification algorithms, such as: more classic genetic algorithms, ant colony algorithms, and particle swarm algorithms. Most of these algorithms are used for parameter optimization of classification algorithms. There is no sample selection for training samples, because the quality of training samples is also the focus of fault diagnosis algorithms, and samples with poor quality greatly affect the classification results of the algorithm, so it is very important to develop a transformer sample selection method

Images